Code: https://gist.github.com/thistleknot/22a3ee8fe20c3e12504d72a61f4d6ebc

I finished version 3 of my text to graph model.

I used spacy with manually labelled data (using Universal Data Tool) to train two models, one for entities, and one for relations. I then predict on a new dataset.

I used up my monthly limit on chatGPT to work up to this solution, but it’s there. And I have the perfect use case for it.

Now I should be able to build knowledge graphs from text data using Subject – modifier/action – object entities and relation extractor. In fact, I think I should upload my model to huggingface.